Product Update: Experiment Health Score
At Effective Experiments, we help our customers build trustworthy experimentation programs. Our firm belief is that Experimentation programs built on strong foundations of Governance, Process and Structure will outlast the programs that just run experiment after experiment but get nowhere in the long run.
The challenge with organizations trying to grow their experimentation program across the wider business is it faces an uphill challenge to get a disparate group of people to run experiments at the same level of quality.
Often, there is no visibility of the quality of the experiments run. We talked about this in our blog post on Experimentation blindspots.
We are proud to release a new feature called The Experiment Health Score. It’s the first in our series of Experimentation Program Scorecard built on top of our Experimentation Ops framework.
The automated Experiment Health Score provides the program orchestrators with visibility on the quality of the experiment pre-live and post-completion.
When an experiment is yet to go live, the Experiment Health Score will provide visibility on how well an experiment is planned.
- Does it have a hypothesis in place in the ideation stage or later?
- Does it have well defined metrics
- Has the idea come from research or on the fly?
- Has the experiment followed a well defined process
- Is there a business initiative or is it just a random experiment
When the experiment has finished, it will look for
- The presence of learnings and action steps
- Clear reporting of metrics. It will look out for retrospectively changed details.
Experiment scores are automatically calculated in Effective Experiments and are visible on the individual experiment pages. Our customer success manager and strategic consultants will also show you a breakdown of the Health Score by team and individual so you know where improvements can be made with coaching.
For experimentation programs not using Effective Experiments, you can download a free Google Sheet we have prepared where you can manually score each experiment based on the data you have available from different systems. Whilst this is not automated and won’t guarantee